Rethinking Diverse Human Preference Learning through Principal Component Analysis

Open in new window